Tropical Cyclones and Global Climate Change: A Post-IPCC Assessment

Tropical Cyclones and
Global Climate Change:
A Post-IPCC Assessment
A. Henderson-Sellers,* H. Zhang,+ G. Berz,# K. Emanuel,@ W. Gray,& C. Landsea,**
G. Holland,+ J. Lighthill,++ S-L. Shieh,## P. Webster,@@ and K. McGuffie+

ABSTRACT
The very limited instrumental record makes extensive analyses of the natural variability of global tropical cyclone
activities difficult in most of the tropical cyclone basins. However, in the two regions where reasonably reliable records
exist (the North Atlantic and the western North Pacific), substantial multidecadal variability (particularly for intense Atlantic hurricanes) is found, but there is no clear evidence of long-term trends. Efforts have been initiated to use geological and geomorphological records and analysis of oxygen isotope ratios in rainfall recorded in cave stalactites to establish a
paleoclimate of tropical cyclones, but these have not yet produced definitive results. Recent thermodynamical estimation of the maximum potential intensities (MPI) of tropical cyclones shows good agreement with observations.
Although there are some uncertainties in these MPI approaches, such as their sensitivity to variations in parameters
and failure to include some potentially important interactions such as ocean spray feedbacks, the response of upperoceanic thermal structure, and eye and eyewall dynamics, they do appear to be an objective tool with which to predict
present and future maxima of tropical cyclone intensity. Recent studies indicate the MPI of cyclones will remain the
same or undergo a modest increase of up to 10%–20%. These predicted changes are small compared with the observed
natural variations and fall within the uncertainty range in current studies. Furthermore, the known omissions (ocean spray,
momentum restriction, and possibly also surface to 300-hPa lapse rate changes) could all operate to mitigate the predicted intensification.
A strong caveat must be placed on analysis of results from current GCM simulations of the “tropical-cyclone-like”
vortices. Their realism, and hence prediction skill (and also that of “embedded” mesoscale models), is greatly limited by
the coarse resolution of current GCMs and the failure to capture environmental factors that govern cyclone intensity.
Little, therefore, can be said about the potential changes of the distribution of intensities as opposed to maximum achievable
intensity. Current knowledge and available techniques are too rudimentary for quantitative indications of potential changes
in tropical cyclone frequency.
The broad geographic regions of cyclogenesis and therefore also the regions affected by tropical cyclones are not
expected to change significantly. It is emphasized that the popular belief that the region of cyclogenesis will expand
with the 26°C SST isotherm is a fallacy. The very modest available evidence points to an expectation of little or no
change in global frequency. Regional and local frequencies could change substantially in either direction, because of
the dependence of cyclone genesis and track on other phenomena (e.g., ENSO) that are not yet predictable. Greatly
improved skills from coupled global ocean–atmosphere models are required before improved predictions are possible.

*Chancellory, Royal Melbourne Institute of Technology,

Melbourne, Australia.
+
Mesoscale Meteorology Research Group, BMRC, Melbourne,
Australia.
#
Reinsurance/Research and Development, Munich Reinsurance
Company, Munich, Germany.
@
Center for Meteorology and Physical Oceanography, Massachusetts Institute of Technology, Cambridge, Massachusetts.
&
Department of Atmospheric Sciences, Colorado State University, Fort Collins, Colorado.
**NOAA AOML/Hurricane Research Division, Miami, Florida.
++
Department of Mathematics, University College London, London, United Kingdom.
Bulletin of the American Meteorological Society

##

National Taiwan University/Central Weather Bureau, Taipei,
Taiwan.
@@
Program in Atmospheric and Oceanic Science, University of
Colorado, Boulder, Colorado.
Corresponding author address: Professor A. Henderson-Sellers,
Deputy Vice-Chancellor (Research and Development), Royal
Melbourne Institute of Technology, P.O. Box 71, Bundoora VIC
3083, Australia.
E-mail:
In final form 10 September 1997.
©1998 American Meteorological Society

19


1. Introduction
a. Tropical cyclones
Tropical cyclones are perhaps the most devastating
of natural disasters both because of the loss of human
life they cause and the large economic losses they induce (Anthes 1982; Gray and Landsea 1992; Gray et
al. 1993, 1994; Tonkin et al. 1997; Diaz and Pulwarty
1997). Vulnerability to tropical cyclones is becoming
more pronounced because the fastest population
growth is in tropical coastal regions. Understanding
tropical cyclone genesis, development, and associated
characteristic features has been a challenging subject
in meteorology over the last several decades. In recent
years, attempts to associate tropical cyclone trends
with climate change resulting from greenhouse warming has led to additional attention being paid to tropical cyclone prediction (e.g., Emanuel 1987; Evans
1992; Lighthill et al. 1994). Exploring possible changes
in tropical cyclone activity due to global warming is
not only of theoretical but also of practical importance.
A tropical cyclone (TC) is the generic term for a
nonfrontal synoptic-scale low pressure system originating over tropical or subtropical waters with organized convection and definite cyclonic surface wind
circulation. Tropical cyclones with maximum sustained surface winds of less than 17 m s−1 are generally called “tropical depressions.” Once a tropical cyclone achieves surface wind strengths of at least 17
m s−1 it is typically called a “tropical storm” or “tropical cyclone” and assigned a name. If the surface winds
reach 33 m s−1, the storm is called a “typhoon” (the
northwest Pacific Ocean), a “hurricane” (the North
Atlantic Ocean and the northeast Pacific Ocean), or a
“severe tropical cyclone” (the southwest Pacific Ocean
and southeast Indian Ocean) (Neumann 1993).

Tropical cyclones derive energy primarily from
evaporation from the ocean and the associated condensation in convective clouds concentrated near their
center (Holland 1993), as compared to midlatitude
storms that primarily obtain energy from horizontal
temperature gradients in the atmosphere. Additionally,
tropical cyclones are characterized by a “warm core”
(relatively warmer than the environment at the same
pressure level) in the troposphere. The greatest temperature anomaly generally occurs in the upper troposphere around 250 hPa. It is this unique warm-core
structure within a tropical cyclone that produces very
strong winds near the surface and causes damage to
coastal regions and islands through extreme wind,
storm surge, and wave action.
20

Tropical cyclones occur predominantly over tropical oceans where observed meteorological data are
scarce. In addition, the destructive nature of tropical
cyclones makes their observations difficult and expensive. Reconnaissance aircraft, satellite observations,
radar observations, rawindsonde observations, and
conventional surface observations are used in monitoring tropical cyclone frequency and intensity. The
best method of observing a tropical cyclone is by direct observations from reconnaissance aircraft, particularly for monitoring location and intensity. Satellite
data, although extremely useful and widely used, are
not a complete substitute for reconnaissance aircraft
observations because of the difficulties involved in
translating radiances into required parameters. The
Dvorak technique (Dvorak 1984) in combination with
spiral overlays and subjective interpretations is commonly applied to estimate the location and intensity
of tropical cyclones from satellite imagery. However,
there may be large errors if these estimates are made
from the satellite observations alone, and calibration
procedures based on aircraft reconnaissance have so

far only occurred for the North Atlantic and western
North Pacific. The uncertainties associated with the
satellite imagery analysis are discussed in detail by
Holland (1993). Unfortunately, the high cost of reconnaissance aircraft means that such observations are
now routinely available only in the North Atlantic
Ocean.
Each year approximately 80–90 tropical cyclones
reaching tropical storm intensity occur around the
globe (Gray 1979; Anthes 1982; Frank 1987; McBride
1995) with about two-thirds of these reaching hurricane intensity. The earlier statistics are updated to
1995 in Table 1. The globally averaged annual variation of cyclone occurrence is only about 10%. Regional variations are much larger, often around 30%,
and no obvious correlations exist in variations between different regions (Raper 1993). For instance, in
the Australian–southwest Pacific region, the average
number of tropical cyclones observed during 1950–
86 was 14.8, with an annual variation of 40% (Evans
1990). As pointed out by Holland (1981), the quality
of the tropical cyclone databases can be highly variable. Different definitions, techniques, and observational approaches may produce errors and biases in
these datasets that could have implications for the
study of the natural variation of tropical cyclone activities and the detection of possible historical trends
(e.g., Nicholls et al. 1998, manuscript submitted to J.
Climate).
Vol. 79, No. 1, January 1998


TABLE 1. Averaged annual total numbers of tropical cyclones (wind at least 17 m s−1) and intense tropical cyclones (wind at least
33 m s−1) and their standard deviations over all tropical cyclone basins (unit: number per year). Data are retrieved from National Climate
Data Center GTECCA dataset for the period 1970–95.
North
Atlantic
Basin

Mean

Std

East
North
Pacific
Basin

North
Indian
Basin

Southwest
Indian
Basin

Southwest
Pacific
Basin

West
North
Pacific
Basin

Totals

TC

9.3

17.8

5.2

10.6

16.4

26.8

86.1

Intense TC

5.0

10.3

2.0

4.8

7.5

16.4

45.9

TC

2.6

4.7

2.2

3.2

4.6

3.9

7.9

Intense TC

1.7

3.5

1.9

2.6

2.6

3.4

7.0

Note: Totals = global total numbers.

Gray (1968, 1975) produced a global map of genesis points for all tropical cyclones over the 20-yr period 1952–71. Preferred regions of tropical cyclone
formation include the western Atlantic, eastern Pacific,
western North Pacific, north Indian Ocean, south Indian Ocean, and Australian–southwest Pacific. Most
of the cyclones (87%) formed between 20°N and 20°S.
About two-thirds of all tropical cyclones form in the
Northern Hemisphere, and the number of tropical cyclones occurring in the Eastern Hemisphere is about
twice that in the Western Hemisphere. These differences are partially due to the absence of tropical cyclones in the South Atlantic and the eastern South
Pacific during the 20-yr period study.
Tropical cyclones are seasonal phenomena: most
tropical ocean basins have a maximum frequency of
cyclone formation during the late summer to early
autumn period. This is associated with the period of
maximum sea surface temperature (SST), although
other factors, such as the seasonal variation of the
monsoon trough location, are also important (Frank
1987; McBride 1995). In the Australian region, the
tropical cyclone season typically extends from November to May with maximum cyclone activity in
January and February (e.g., Holland 1984a,b; Holland
et al. 1988; Evans 1990). The storm season in the
North Atlantic becomes highly active during August–
October, with a maximum frequency of occurrence in
September (Neumann et al. 1985). The average tropical cyclone occurrence over the western North Pacific
is about 26 per year, with a maximum cyclone activity in August and a highly seasonal variation. This total
is more than in any other region (Xue and Neumann
1984), and this is also the only region where tropical

Bulletin of the American Meteorological Society

cyclogenesis has been observed in all months of the
year. The western North Pacific is particularly noted
for the occurrence of very large and very intense tropical storms (Frank 1987; McBride 1995). Indeed, the
12 lowest central pressures in the global record have
been observed for the tropical cyclones in the western
North Pacific (Holland 1993).
The favorable locations for tropical cyclone genesis
are in or just poleward of the intertropical convergence
zone (ITCZ) or a monsoon trough (Gray 1968). The
ITCZ is generally located near the monsoon shear line
between low-level equatorial westerlies and easterly
trades. The disturbances embedded in the easterly trade
wind flow are also conducive to the formation of tropical cyclones (Frank 1987).
The physical parameters favorable for cyclogenesis
have been summarized by Gray (1968, 1975, 1979,
1981). He found that the climatological frequency of
tropical cyclone genesis is related to six environmental factors: (i) large values of low-level relative vorticity, (ii) Coriolis parameter (at least a few degrees
poleward of the equator), (iii) weak vertical shear of
the horizontal winds, (iv) high SSTs exceeding 26°C
and a deep thermocline, (v) conditional instability
through a deep atmospheric layer, and (vi) large values of relative humidity in the lower and middle troposphere.
Although the above six parameters are not sufficient
conditions for cyclogenesis, Gray (1975, 1981) argued
that tropical cyclone formation will be most frequent
in the regions and seasons when the product of the six
genesis parameters is a maximum. Gray defined the
product of (i), (ii), and (iii) as the dynamic potential
for cyclone development, and the product of (iv), (v),

21


counted; and (iv) natural variability of tropical
storms is very large, so small trends are likely
to be lost in the noise. (p. 334)

and (vi) may be taken as the thermodynamic potential. He derived the seasonal genesis parameter from
these six parameters (Gray 1975).
b. Tropical cyclones and climate change
The Intergovernmental Panel on Climate Change
(IPCC) “Impacts, Adaptation and Mitigation of Climate Change” report (Watson et al. 1996) stated that
Reinsurers have noted a fourfold increase in disasters since the 1960s. This is not due merely
to better recording, because the major disasters—which account for 90% of the losses and
would always be recorded—have increased just
as quickly. Much of the rise is due to socioeconomic factors, but many insurers feel that the
frequency of extreme events also has increased.
(p. 547)

It also stated that
Insurers had at least one “billion dollar” storm
event every year from 1987 to 1993. With such
an unexpectedly high frequency, some local
insurance companies collapsed, and the international reinsurance market went into shock.
(p. 547)

This IPCC report went on to note that
Traditionally, insurers have dealt with changes
in risk in four ways: restricting coverage so that
the balance of risk-sharing shifts toward the

insured; transferring risk; physical risk management (before and after the event); or raising
premiums. However, in view of the increasing
costs of weather claims, insurers now are considering a more fundamental approach. . . . Lack
of information about extreme events hampers
such activity and makes insurers wary of committing their capital. (p. 548)

At the same time, the IPCC “Science of Climate
Change” report (Houghton et al. 1996) stated that
the-state-of-the-science [tropical cyclone simulations in greenhouse conditions] remains poor
because (i) tropical cyclones cannot be adequately simulated in present GCMs; (ii) some
aspects of ENSO are not well simulated in
GCMs; (iii) other large-scale changes in the
atmospheric general circulation which could
affect tropical cyclones cannot yet be dis22

and
In conclusion, it is not possible to say whether
the frequency, area of occurrence, time of occurrence, mean intensity or maximum intensity
of tropical cyclones will change. (p. 334)

Research efforts focused on assessing the potential
for changes in tropical cyclone activity in the greenhouse-warmed climate have progressed since those
that were the basis of this IPCC assessment (themselves undertaken in 1994 and early 1995). This paper synthesizes the input from the members of a steering committee of the World Meteorological Organization Commission for Atmospheric Sciences and reflects recent experimental results in a summary of the
new findings in this field.
This review should be read in the context of our
current situation with regard to tropical cyclone predictions for a greenhouse-warmed world.
1) WHAT DO WE KNOW?
1) Tropical cyclones are currently devastatingly severe weather events.
2) Human vulnerability to TCs is increasing because
of increasing populations on tropical coasts.

3) Tropical cyclone formation and intensity change
are currently very difficult to predict.
4) Costs of TC impacts are increasing because of increasing costs of infrastructure and increasing “responsibility” claims on private and public funds.
5) The balance of evidence indicates that greenhouse
gas emissions are producing climate change
(Houghton et al. 1996).
6) Concern about possible future changes in tropical
cyclone activity relates to changes in (i) frequency
of occurrence, (ii) area of occurrence, (iii) mean intensity, (iv) maximum intensity, and (v) rain and
wind structure.
2) WHAT DO WE NOT KNOW?
1) How to predict TCs today: genesis, maximum intensity.
2) How the environmental parameters that appear to
be important for TC genesis will change.
3) How the large-scale circulation features that appear
Vol. 79, No. 1, January 1998


to be linked to TC climatology, especially the
quasi-biennial oscillation (QBO) and El Niño–
Southern Oscillation (ENSO), will change.
4) How the upper-ocean thermal structure, which acts
as the energy source for TC development, will
change.
3) WHAT TOOLS ARE AVAILABLE TO US NOW?
1) Coupled ocean–atmosphere general circulation
models. (OAGCMs). These are providing useful
information on the general characteristics of climate change, but they currently have coarse resolution (about 500 km), climate drift (or are energy
corrected), and unproven skill for present-day TCs.
2) Atmospheric general circulation models (AGCMs)

linked to mixed layer ocean submodels or employing SST predictions from OAGCMs have better
resolution (about 100 km) but are still too coarse
for mesoscale dynamics and share the latter two
drawbacks of OAGCMs (as in 1).
3) Mesoscale models driven off-line from the output
of OAGCMs or AGCMs have better resolution
(about 20 km) but still share the other drawbacks
(as for 1 and 2).
4) Empirical relationships such as Gray’s genesis parameters or (much too?) simply SSTs alone suffer
from drawbacks associated with empiricism.
5) “Upscaling” thermodynamic models, such as those
of Emanuel (1991) and Holland (1997), are known
not to capture all processes of importance.
This review is phrased in terms of doubled CO2 climate conditions for simplicity and because the evaluations assessed predate any attempt to consider the
additional impacts of sulfate, or other, aerosols on
tropical cyclones. However, all the assessments are
equally applicable to greenhouse conditions modified
by either or both other greenhouse gases or atmospheric aerosols.
2. Natural variability in tropical
cyclones and possible trends
Ascertaining tropical cyclone variability on
interannual to interdecadal timescales is hampered by
the relatively short period over which accurate records
are available. For the North Atlantic Basin (including
the North Atlantic Ocean, the Gulf of Mexico, and the
Caribbean Sea), aircraft reconnaissance has helped to
Bulletin of the American Meteorological Society

provide a nearly complete record since the mid-1940s.
The western North Pacific Basin (i.e., the Pacific north

of the equator and west of the dateline, including the
South China Sea) also has had extensive aircraft surveillance giving high quality records since the mid1940s. For the remaining tropical cyclone areas (the
north Indian, the southwest Indian, the Australian–
southeast Indian, the Australian–South Pacific, and the
northeast Pacific Oceans), there are only about 25–30
years of reliable measurements of annual activity derived from satellites. Thus, with the instrumental
record so limited, it is difficult to make persuasive
analyses of trends and of the physical mechanisms
responsible for tropical cyclone variability.
However, even with these limitations, some information can be established about tropical cyclones in
the past. The averages and standard deviations over
these last couple of decades for each tropical cyclone
area are well established (e.g., Neumann 1993). While
the North Atlantic Basin averages 9–10 tropical cyclones reaching tropical storm strength (winds at least
17 m s−1) of which 5–6 reach hurricane strength (winds
at least 33 m s−1), these compose only about 12% of
the world total. The most active region, globally, is the
western North Pacific with an annual average of 26
tropical storms and, of these, 16 typhoons (winds of
at least 33 m s−1), composing over 30% of the world
total. Overall, the number of tropical cyclones reaching 17 m s−1 averages 84 globally, with a range of plus/
minus one standard deviation from 76 to 92. Hurricane-force tropical cyclones average 45 each year with
a range of plus/minus one standard deviation from 39
to 51.
Among the basins with only relatively short reliable records, Nicholls (1992) identified a downward
trend in the numbers of tropical cyclones occurring in
the Australian region from 105°–165°E, primarily
from the mid-1980s onward. However, it is likely that
this change is primarily artificial, due to changes in
tropical cyclone analysis procedures (Nicholls et al.

1998, manuscript submitted to J. Climate). As shown
in Fig. 1a, if only more intense tropical cyclones are
counted (i.e., those with a minimum pressure of less
than 990 hPa) much of the downward trend in cyclone
numbers is removed. In the remaining short-record
basins, the northeast Pacific has experienced a notable
upward trend, the north Indian a notable downward
trend, and no appreciable long-term variation is observed in the southwest Indian and southwest Pacific
(east of 165°E) based upon data from the late 1960s
onward (adapted from Neumann 1993). However,
23


(a)

(b)

FIG. 1. (a) Time series of the number of tropical cyclones in the Australian region (105°–165°E) between 1969 and 1995. Dark
bars indicate the number of cyclones with minimum pressure below 990 hPa. The gray bars indicate the numbers with minimum
pressure between 1000 and 990 hPa [adapted from Nicholls et al. (1998, manuscript submitted to J. Climate)]. (b) Time series of
Atlantic basin intense hurricanes (dark bars) and weaker cyclones (gray bars) for 1944–96. Intense hurricanes are those cyclones that
attain sustained surface winds of at least 50 m s−1 at some point in their life cycle. Weaker cyclones include all other remaining tropical
storms, subtropical storms, and hurricanes. The superimposed lines are the linear best fits for the intense hurricanes (lower line) and
for the total number of cyclones (upper line) [from Landsea et al. (1996)].

whether these represent longer-term changes or reflect
shorter-term (on the scale of tens of years) variability
is completely unknown because of the lack of long,
reliable records.
For the northwest Pacific basin, Chan and Shi

(1996) found that both the numbers of typhoons and
the total number of tropical storms have been increasing since about 1980. However, the increase was preceded by a nearly identical decrease from about 1960
to 1980. No analysis has been undertaken as yet of the
numbers of intense typhoons (winds at least 50 m s−1)
because of an overestimation bias in the intensity of
such storms in the 1960s and 1970s (Black 1993).
There has been an extensive analysis for the Atlantic basin in part because of the length of the reliable
record for this basin (back to 1944) and for U.S. coastal
landfalling hurricanes (back to 1899). In common with
the northwest Pacific data, observations for this basin
also have a bias in the measurement of strong hurricanes: from the 1940s through to the 1960s, the intensity of strong hurricanes is believed to have been overestimated by 2.5–5 m s−1 (Landsea 1993). This bias has
been crudely removed to provide estimates of the true
occurrence of intense (or major) hurricanes, those with
winds of at least 50 m s−1, which are designated as a
category 3, 4, or 5 on the Saffir–Simpson hurricane
intensity scale (Simpson 1974).
24

Examination of the record of the number of Atlantic tropical storms (including those designated as subtropical storms from 1968 onward) shows substantial
year-to-year variability, but no significant trend
(Landsea et al. 1996) (Fig. 1b). In contrast, intense
hurricanes exhibited a pronounced downward trend
from the 1940s through the 1990s, despite the near
record-breaking years of 1995–96. In addition to these
changes in frequency, there has been a decrease in the
mean intensity of the Atlantic tropical cyclones, although there has been no significant change in the
peak intensity reached by the strongest hurricane each
year. Fluctuations in numbers of intense hurricanes are
considerable in the 1940s through to the late 1960s;
although there is a period of reduced activity in the

1970s through 1994, and a “spike” of activity in 1995
(Fig. 1b).
These trends for the entire Atlantic basin are mirrored by those from intense hurricanes striking the
U.S. east coast, from the peninsula of Florida to New
England (Landsea 1993). The quiet period of the
1970s to the early 1990s is similar to a quiescent regime in the first two decades of this century. Active
conditions began in the late 1910s and continued into
the 1960s. During two particularly active periods, the
Florida peninsula and the upper Atlantic coast (from
Georgia to New England) experienced seven intense
Vol. 79, No. 1, January 1998


hurricane landfalls in two periods each of seven years
(1944–50 and 1954–60). Other time series for all U.S.
hurricanes (Hebert et al. 1996) and for hurricanes affecting land areas around the Caribbean also show
some quasiperiodicity.
It has been suggested that the Atlantic’s hyperactive hurricane seasons of 1995 (19 tropical storms, 11
hurricanes, and 5 intense hurricanes) and 1996
(13 tropical cyclones, 9 hurricanes, and 6 intense
hurricanes) may be heralding a return to an active regime similar to that seen between the 1940s and the
1960s (Landsea et al. 1996). Since the Atlantic hurricane activity observed during the 1970s and into the
early 1990s was anomalously low compared with previous decades, a return to a more active regime is not
surprising.
Data for typhoon activities around the island of
Taiwan have been gathered since 1897 and can be used
as an indication of interannual variabilities of the
nonrecurved western North Pacific typhoons defined
as tropical cyclones achieving wind speeds of 33 m s−1
and greater. Recent studies (Chang 1996) focus on the

typhoons that have caused loss of lives and/or damage to properties on the island of Taiwan, regardless
of whether they crossed the coast. In general, there are
about 3–4 such typhoons per year, but there is a pronounced variation from as many as eight (1914) to as
few as zero (1941, 1964). A slight decreasing trend is
apparent, from 4 to 3 typhoons per year, but this may
be a result of change in the definitions of such typhoons in 1962.
The global cyclone frequency taken from the National Climatic Data Center Global Tropical Cyclone
Data Set indicates that the number of tropical cyclones
may have increased since 1970. However, this increase
has arisen entirely from the more poorly observed regions of the Southern Hemisphere and the eastern
North Pacific and cannot be differentiated from changing observing practices and slow, multidecadal oscillations in cyclone numbers.
In the past few years, several attempts have been
initiated aimed at trying to use geological records to
quantify tropical cyclone activity back as far as the end
of the last glacial episode, about 10 000 years ago.
These methods, although still in their infancy, suggest
that there may be potential for quantitative analysis of
changing cyclone characteristics with climate. However, there is insufficient information available at
present for quantitative estimation of trends and natural variability over geological timescales.

Bulletin of the American Meteorological Society

3. Tropical cyclone genesis and
frequency and their potential to
change in greenhouse conditions
The processes that are responsible for development
of tropical cyclones are poorly understood, in large part
because of the lack of good observations of the highly
transient changes that occur. Even in current operational weather forecasts, prediction of tropical cyclone
formation still lacks skill and such forecasting is reduced to “watchful waiting” (Holland 1993), relying

on detecting the satellite signature, combined with
knowledge of current environmental factors and the
genesis climatology of tropical cyclones in the area.
Understanding how tropical cyclone genesis may
change in the greenhouse-warmed climate is certainly
a significant challenge to current research.
The problem of predicting how tropical cyclone
frequency might respond to greenhouse-induced climate change can be broken into two parts: predicting
how the environmental capacity to sustain tropical
cyclones may change and predicting how the frequency and strength of initiating disturbances may
change. The thermodynamic analysis by Holland
(1997) indicates that there could be an enhanced environment for tropical cyclone intensification. GCM
predictions also indicate that the strength of very largescale tropical circulations such as monsoons and the
trade winds are expected to be increased, which could
be expected to provide both an enhanced environment
and more initiating disturbances. Balanced against this
is the predicted increase of upper-tropospheric wind
shear. Substantial uncertainties also exist in known
regional factors correlated with cyclone frequency,
such as Sahel rainfall (Landsea and Gray 1992) or
ENSO (Nicholls 1984). Elementary applications of
empirical relationships from the current climate to a
future climate are fraught with danger and offer little
useful insight.
Gray (1968, 1979) summarized the knowledge of
large-scale conditions necessary for tropical cyclone
genesis, but these are by no means sufficient. The Gray
genesis parameter was applied to GCM results for climate change by Ryan et al. (1992). Their results were
inconclusive and there remains doubt whether such
parameters, which have been highly tuned to fit the

current climate, are directly applicable to changed climate conditions. For example, a widespread misconception is that were the area enclosed by the 26°C SST
isotherm to increase, so too would the area experiencing tropical cyclogenesis. Application of a thermody25


FIG. 2. Plots of MPI of tropical cyclones against sea surface temperatures. (a) MPI estimated from monthly mean atmospheric
temperature soundings and ocean temperatures at several tropical cyclone sites (indicated by different symbols) and compared with
an empirical curve for North Atlantic modified from DeMaria and Kaplan (1994) as discussed in Holland (1997). (b) MPI estimated
from greenhouse conditions constructed by adding the MECCA model simulated atmospheric and oceanic temperature changes near
the tropical cyclone sites in (a) to the observed monthly mean temperature soundings at these sites. Rhomboidal symbols indicate
these estimations at the different sites (not distinguished) and the solid thick line is the best fit curve to these points. Thin dashed lines
indicate the sense of MPI change and thick vertical dash lines show the changing cyclogenesis limit.

namic technique (Holland 1997) to climate change
scenarios in Figs. 2a and 2b clearly indicates that cyclone development in a warmer climate occurs at
higher oceanic temperatures, particularly for intense
tropical cyclones. This arises from upper-atmospheric
warming that compensates to some extent for the increased energy potential from the warmer oceans. This
conclusion is supported by the known importance of
dynamical processes, such as development of a broad
region of upward ascent, which are, themselves, governed by unchanging parameters such as the earth’s
rotation. The finding also concurs with the modeling
studies of Bengtsson et al. (1996). The net result, therefore, is that current knowledge indicates that the broad
geographic regions affected by tropical cyclones are
not expected to change significantly. In particular,
there is no reason to believe that the region of cyclogenesis will expand with the 26°C isotherm.
It is conceivable, however, that changes in the largescale circulation of the atmosphere could increase or
decrease the rate of movement of tropical cyclones out
of their genesis regions and into higher latitudes. It is
also possible that the extratropical transition of tropical cyclones may change in character.
a. Relationships between tropical cyclones and

large-scale circulations
Gray (1984a,b) related tropical cyclone activity in
the western Atlantic and western North Pacific to the
phase of the stratospheric QBO. He found that when
QBO winds are from a westerly direction, there are
26

nearly twice as many hurricane days in the western
Atlantic as compared to when the QBO is in an easterly phase. In westerly phase QBO seasons, there seem
to be more intense and longer lasting Atlantic hurricanes. In the western North Pacific, this relationship
is not as strong as in the western Atlantic. In contrast
to the Atlantic, in the western North Pacific tropical
cyclone activity is more frequent in an easterly phase
of the stratospheric QBO.
In recent years, the ENSO influences on tropical
cyclone activity have been investigated (e.g., Nicholls
1984, 1992; Gray 1984a,b; Revell and Goulter 1986;
Dong 1988; Evans and Allan 1992). El Niño events
have been shown to be related to the seasonal frequency and interannual variations of tropical cyclone
occurrence. Nicholls (1984), Chan (1985), and Dong
and Holland (1994) have clearly shown strong relationships between the ENSO and longitudinal shifts
in the regions of cyclone development.
During an ENSO warm event in the eastern Pacific,
the SSTs over the western Pacific are relatively cooler
and atmospheric pressure over Australia is higher than
normal. This leads to a reduced number of cyclones
in the Australian region (Nicholls 1984; Evans and
Allan 1992), while the center of tropical cyclone activity moves farther east and north (toward the equator), and the frequency of cyclone formation east of
170°E actually increases (Revell and Goulter 1986;
Evans and Allan 1992). In cold ENSO events (La

Niña) these trends are reversed. Nicholls (1992) has
shown that the number of tropical cyclones around
Vol. 79, No. 1, January 1998


Australia (105°–165°E) has decreased rather dramatically since the mid-1980s. Some of this reduction may
be associated with there being more El Niño events
since that time (i.e., 1982–83, 1986–87, 1991–92).
The relationship between the ENSO events and
tropical cyclone activity in the northwest Pacific has
also been studied (e.g., Chan 1985; Dong 1988; Lander
1994). Consistent with results for the Australian–
southwest Pacific basin, there are reduced numbers of
tropical cyclone genesis west of 160°E, but increased
cyclogenesis events in a region east of 160°E and south
of 20°N during El Niño events (Chan 1985; Lander
1994). The opposite occurs during cold ENSO events.
There is also a tendency for the tropical cyclones to
form closer to the equator during El Niño events.
In the North Atlantic region, the ENSO influence
on tropical cyclone activity is quite different (Gray
1984a; Shapiro 1987; McBride 1995). During El Niño
events (ENSO warm phase), tropospheric vertical
shear is increased by the stronger upper-tropospheric
westerly winds. This inhibits tropical cyclone genesis
and intensification. In contrast, seasonal frequency of
tropical cyclone occurrence is slightly enhanced in
non–El Niño years. Unfortunately, since current climate models do not adequately simulate ENSO events,
no definite statements can be made about the likely
impacts of changing climate on these coupled phenomena.

Recently, Landsea and Gray (1992) have detected
a strong empirical relationship between North Atlantic hurricanes and various observable parameters of the
climate system, such as the extent of summer rain in
the Sahel. As yet unpublished work by Holland and
collaborators also indicates that cyclone frequency in
the Australian region may be related to establishment
of suitable thermodynamic preconditions. Holland
(1995) has also hypothesised that large-scale circulation patterns in the western North Pacific are associated with and conducive to extended periods of repeated cyclogenesis. None of these results have yet
been applied directly to cyclone development and climate change.
An interesting, and potentially useful, statistic on
tropical cyclone frequency is that the global frequency
is highly stable from year to year: variations are typically around 10%. This compares markedly with local regional variations that are typically 100% of the
long-term mean (e.g., Fig. 1) and can be more than
200%. It is concluded that current knowledge and
available techniques are too rudimentary for quantitative indications of potential changes in tropical cyBulletin of the American Meteorological Society

clone frequency. However, the available evidence
strongly points to an expectation of little or no change
in global frequency. Regional frequency could change
substantially in either direction.
b. GCM studies of numbers of tropical cyclones
GCMs have been used by a number of groups to try
to infer changes in tropical cyclone activity by analyzing the resolvable-scale vortices that develop. These
studies are subject to a number of caveats and produce
conflicting results: Haarsma et al. (1992) found an
increase in frequency of tropical cyclones, Bengtsson
et al. (1996) found large decreases, and Broccoli and
Manabe (1990) found that increases or decreases could
be obtained by reasonable variations in the model
physics. A commentary on these simulations is provided below.

The possible changes in tropical cyclone activity
associated with greenhouse-induced climate change
have been investigated using GCM results directly
(e.g., Broccoli and Manabe 1990; Haarsma et al. 1993;
Bengtsson et al. 1995, 1996). Broccoli and Manabe
(1990) used the Geophysical Fluid Dynamics Laboratory GCM to study the response of tropical cyclones
to increases in atmospheric CO2. Two versions of the
model, R15 (4.5° lat × 7.5° long) and R30 (2.25° lat ×
3.75° long), were utilized. The cloud treatments
adopted were with fixed cloud and variable cloud
amounts. In the experiments with fixed cloud, the
number and duration of tropical storms increased in a
doubled CO2 climate for the R15 integration. However, a significant reduction of the number and duration was indicated in the experiments with variable
cloud. The response of the simulated number of storms
to a doubling of CO2 is apparently insensitive to the
model resolution but crucially dependent on the parameterization of clouds (Broccoli and Manabe 1992).
Haarsma et al. (1993) undertook similar experiments for present-day and doubled CO2 concentrations. Their model resolution was R30 (2.25° lat ×
3.75° long) with variable cloud amount. Evans (1992)
argued that it is important to examine the physical
mechanisms involved in the generation of the model
“storm” and test the degree to which the model vortices have physical similarities with real tropical cyclones. The simulated tropical disturbances for the
present climate analyzed by Haarsma et al. (1993) have
a much larger horizontal extent and weaker intensity
than those observed, but some physical features of
tropical cyclones, such as low-level convergence, upper-tropospheric outflow, and a warm core, were pro27


duced by this GCM. In the doubled CO2 conditions,
the number of simulated tropical storms increases by
about 50%.

Bengtsson et al. (1995, 1996) investigated the influence of greenhouse warming on tropical storm climatology, using a high-resolution GCM at T106 resolution (triangular truncation at wavenumber 106,
equivalent to 1.1° lat–long). Their studies suggest a
substantial reduction in the number of storms, particularly in the Southern Hemisphere. They attribute this
reduction to a warming in the upper troposphere, enhanced vertical wind shear, and other large-scale
changes in the tropical circulation such as reduced lowlevel relative vorticity. In comparison to the results for
the control experiment, there are no changes in the
geographical distribution of the GCM-simulated
storms. The seasonal variability of the storm distribution is said to be in agreement with that of the present
climate. However, application of their model results
may be limited by their model’s sensitivity to its resolution and perhaps also by incompatibilities in the experiment. In the ECHAM3 (T106) doubled CO2 experiment (Bengtsson et al. 1996), the fixed global
SSTs were taken from ECHAM3 (T21) doubled CO2
experiment of Cubasch et al. (1992) in which an enhanced tropical hydrological cycle by a strengthened
ITCZ was simulated with a fully coupled ocean model
and the SSTs were warmed between 0.5° and 1.5°C.
Surprisingly, with such high global SSTs and noting
the results from the underpinning experiment,
Bengtsson et al. (1996) reported a weakened tropical
hydrological cycle in their high-resolution experiment.
This weakening in tropical circulation appeared to be
one of the primary reasons for the decrease in the
model’s tropical cyclone activity. It appears that this
model’s tropical climate is very sensitive to its horizontal resolution. It is possible that the changes of
SSTs in the doubled CO2 climate, if simulated by the
high-resolution AGCM coupled with the same OGCM
as Cubasch et al. (1992), may be different from the
ones used in Bengtsson et al. (1996) and thus might
give a different prediction of the changes in tropical
cyclone activities in the greenhouse-warmed climate.
An alternative approach to prediction of the potential changes in tropical cyclone activity is to apply
Gray’s (1968, 1975) seasonal genesis parameter to

GCM fields (e.g., Ryan et al. 1992). A recognized
weakness of this is that the Gray genesis parameter was
derived based on the present climate, but it does not
account for how well the parameter would govern
tropical cyclogenesis in a different climate (Tonkin et
28

al. 1997). Watterson et al. (1995) used Gray’s seasonal
genesis parameters as an objective criterion to derive
a model’s climatology of tropical cyclone genesis from
a GCM with 3.2° lat × 5.6° long resolution. First they
applied the genesis parameters to the European Centre for Medium-Range Weather Forecasts climatology
and compared the results with the observed cyclogenesis. Although their results confirmed the success of
these genesis parameters as a diagnostic tool for locating the genesis regions of tropical cyclones, they
found that these parameters overestimated the number of tropical cyclones in the Southern Hemisphere.
Results from the GCM climatology also show the sensitivity of the model tropical cyclogenesis to the SST
variations and that Gray’s seasonal genesis parameters
have deficiencies in diagnosing both climatological
and interannual tropical cyclone frequency.
Recently, Walsh and Watterson (1997) studied the
tropical-cyclone-like vortices in a limited area model
focused on the Australian continent and nested into a
GCM. This limited-area model has a horizontal resolution of 125 km and has successfully simulated some
of the physical features of tropical cyclones such as
the warm core, low-level wind maxima in the modeled tropical-cyclone-like vortices detected using objective genesis parameters. Compared with observed
cyclogenesis over these regions, this study showed that
although Gray’s seasonal genesis parameters have
some skill in predicting model cyclogenesis for current climate conditions, it is not a definitive measure
and a reformulation of such parameters may be warranted. Walsh and Watterson (1997) identified two
main limitations of climate models that constrain the

model capability for simulating small and convectivedriven systems such as tropical cyclones: coarse horizontal and vertical resolutions and inadequate representation of moist convective processes.
4. Tropical cyclone intensities and their
potential to change in greenhouse
conditions
The sensitivity of tropical cyclone intensity to SST
change has been investigated using a variety of numerical modeling techniques (e.g., Baik et al. 1990;
Drury and Evans 1993; Evans 1993; Bengtsson et al.
1994). With an axisymmetric tropical cyclone model,
Baik et al. (1990) performed extensive sensitivity experiments. They found that when only the SST was
varied, the intensity of the model-simulated storm inVol. 79, No. 1, January 1998


creased with warmer SST and decreased with cooler
SST. Considering the impacts of moist convective
instability on tropical cyclone intensity, Drury and
Evans (1993) explored the sensitivity of a simulated
storm to increased SST and demonstrated that there
seems to be the potential for more intense, wetter tropical cyclones in a moister and warmer world. In their
experiments, the atmosphere with warmer SST was
adjusted such that the convective available potential
energy (CAPE) of the lower-tropospheric air was unchanged. They found that the changes in simulated
cyclone intensity are significantly less than those in
the experiment in which only SST was varied. It
should be mentioned that previous studies of the sensitivity of maximum potential intensity (e.g., Emanuel
1986) also held CAPE fixed while SST was varied.
Historical data covering five tropical ocean basins
for the 20-yr period 1967–86 were examined by Evans
(1993) to identify the relative importance of SST in
the tropical cyclone intensification process. The results
indicate that while SST does influence tropical cyclone

development and provides an upper bound on tropical storm intensity, it is not the overriding factor in
determining the maximum intensity attained by a
storm. Based on empirical evidence, McBride (1981)
found that SST does not seem to be the primary variable in determining whether incipient storms develop,
although a warm ocean surface is needed for tropical
cyclone formation and development.
Evans et al. (1994) utilized a limited area model to
study potential changes in tropical cyclone intensity
to varying SSTs. They performed several experiments
for two well-observed tropical cyclones occurring simultaneously in the northern Australian region. The
sensitivity studies reveal that if the underlying SST is
warmed, the minimum central pressure will decrease
and the associated rainfall will increase. In other
words, tropical cyclones could become stronger than
in the current climate with warmer SSTs, if other environmental conditions are held constant.
a. Thermodynamic model studies of tropical cyclone
intensity
The maximum intensity that can be reached by
tropical cyclones is ultimately limited by the available
energy in the atmosphere and ocean. It has been well
established (e.g., Byers 1944; Riehl 1954; Malkus and
Riehl 1960) that the atmosphere alone cannot provide
sufficient energy for the development of a very intense
tropical cyclone. The warm tropical oceans support
intense cyclone development by a feedback process in
Bulletin of the American Meteorological Society

which falling surface pressures in the cyclone core
release additional energy from the ocean surface. Adverse atmospheric conditions, together with internal
cyclone dynamics and local oceanic cooling by mixing and upwelling, often prevent tropical cyclones

from achieving this theoretical limit (Holland 1997).
While the internal dynamics of tropical cyclones
and the manner in which they interact with their environment are extremely complex and not well understood, the maximum potential intensity (MPI) has been
estimated in recent years by a consideration of the
energetics (e.g., Kleinschmidt 1951; Emanuel 1986,
1991; Holland 1997). Although these techniques involve a number of simplifying assumptions and caveats, Tonkin (1996) has shown that the techniques of
both Emanuel (1991) and Holland (1997) exhibit considerable skill when evaluated using monthly mean
and daily soundings from a large number of stations
in the western Pacific and North Atlantic Oceans. Figure 3 (for the Australian–southwest Pacific region)
shows that these two techniques provide an MPI bound
on the climatological record, recognizing that the short
cyclone record will not include all possible combinations of extreme cases. Furthermore, Fig. 4 indicates
that there is substantial potential skill in forecasting
the maximum intensity of individual cyclones using
observations of ambient atmospheric and oceanic conditions.
Application of the Holland (1997) technique to
current and future climate conditions is illustrated in
Fig. 2. MPI estimates made from monthly mean atmospheric temperature soundings and oceanic temperatures (SST) at several tropical radiosonde sites in the
Northern and Southern Hemispheres were first compared with tropical cyclone observations. Because the
atmospheric conditions are closely tied to the surface
temperatures of tropical oceans, a plot of SST versus
MPI provides a convenient display of the results (Fig.
2a). The theoretical estimates agree closely with the
observed curve of worst-case tropical cyclones for
warm oceans and accurately reproduce the well-known
requirement of SST >26°C for cyclone development
(Gray 1968). The scatter of MPI near 26°C is partially
due to the method and partially reflects real changes
of MPI–SST relationships between ocean basins (e.g.,
Evans 1993). At cooler SSTs, the observations are

composed of cyclones that developed over warmer
tropical oceans and are decaying as they move
poleward. The sensitivity of the model’s estimations
to a variety of parameters employed in such approaches is discussed in detail by Holland (1997).
29


FIG. 3. Seasonal march of MPI predicted by applying the techniques of Holland (1997) and Emanuel (1991) to a number of monthly
mean radiosonde soundings in the Australian–southwest Pacific region, together with the observed extreme cyclone intensities (shaded
region). The solid line is the ambient surface pressure; the line with dots is the MPI from Emanuel model; the dashed line is the MPI
from Holland model [adapted from Tonkin (1996)].
30

Vol. 79, No. 1, January 1998


Taking the climate model forecasts of monthly
mean atmospheric and oceanic temperature changes
at Willis Island and Guam radiosonde sites from those
MECCA (Model Evaluation Consortium for Climate
Assessment) GCMs that have adequately good simulations of observed climate and adding these to the
observed temperatures used in Fig. 2a results in Fig.
2b. The minimum SST at which tropical cyclones develop increases by 2°–3°C. This is similar to that
modeled by GCMs for the ocean changes in a warmed
climate. Thus, the geographical region of cyclogenesis
will remain roughly unchanged. A small increase of
cyclone intensity is predicted, consistent with the more
comprehensive examination by Li (1996). This increase of MPI is reasonably independent of the choice
of parameter values used in the thermodynamic technique, provided that no unforeseen changes occur in
these parameter values with climate change.

The thermodynamic approaches provide an objective estimate of the lowest central pressure that can be
achieved. This provides a conservative parameter for
indicating current and potential future changes in cyclone intensity, and is in accord with the archiving
practices of most major cyclone centers. There are
direct relationships between central pressure and maximum winds (e.g., Holland 1993, chap. 9), but a number of factors, including asymmetries from the cyclone
motion and local wind transients, result in a significant
scatter. We note that the maximum winds vary as roughly
the square root of the central pressure, so that percentages in expected wind changes will be slightly less than
the percentage changes predicted for central pressure.
There is considerable sensitivity to choice of, or
variations in, some parameters used in these thermodynamic models. For example, Holland (1997) assumes a value of 90% for the relative humidity under
the eyewall. This value is consistent with existing information on the conditions in this region, and it produces satisfactory predictions for current climate
(Tonkin 1996). However, its use grossly oversimplifies the complex interactions between wind, ocean, and
spray occurring in the maximum wind region. Several
studies have shown that the presence of spray considerably modifies the near-surface atmospheric layer for
winds above 20 m s−1 (Pudov and Petrichenko 1988;
Fairall et al. 1994), but virtually nothing is understood
of the effects at very high wind speeds.
Recently, J. Lighthill (1996, personal communication) proposed a mechanism relating spray to the thermodynamics of tropical cyclones, based upon the work
of Fairall et al. (1994), who undertook a fluid-dynamiBulletin of the American Meteorological Society

cal analysis of extensive observations made at sea in
winds as high as 25–30 m s−1. These observations
found a substantial layer of “a third fluid” (ocean
spray) between the atmosphere and the ocean and
measured wind temperatures were substantially less
than the SST. The existence of such a temperature
shortfall would affect the saturated water-vapor concentration and therefore the maximum latent-heat content of the air around the tropical cyclone eyewall base.
Moreover, an extrapolation by Fairall et al. (1994) to
40 m s−1 wind speed suggests that the mass density of

spray might rise to only 0.008 kg m−3 (less than 1% of
the air density) and yet that vapor transfer from spray
to air could exceed direct transfer from the ocean surface by an order of magnitude. It is, therefore, proposed
that there is a need for a modest correction to established views of tropical cyclone thermodynamics.
Specifically, if much of the vapor transfer to surface
winds came from spray droplets, then cooling from the
corresponding latent heat transfer might not be fully
compensated by sensible-heat transfer from the ocean
surface, so that air temperature (as observed) would
reach an equilibrium value below that of the ocean
surface. The consequence would be that the average
temperature of saturated air around the base of the
eyewall would be less than the SST. From the thermodynamic viewpoint, the importance of such a correction to the temperature of saturated air around the base
of the eyewall stems from the associated very substantial reduction in latent heat intake per unit mass of air,
consequent on the very steep dependence of saturated
water-vapor concentration on temperature. This type
of mechanism has not yet been incorporated in the
current MPI models.
Since a primary mechanism for tropical cyclone
intensity is the balance between input of mechanical
energy from buoyancy forces acting on saturated air
rising in the eyewall (approximately, along a moistair adiabat) and dissipation of wind energy in the turbulent boundary layer at the ocean surface, then ocean
spray may provide a self-limiting process. The energy
input per unit mass of air must be reduced if air at the
base of the eyewall has a water-vapor concentration
well below that associated with the SST, whereas dissipation in the turbulent boundary layer is unlikely to
be greatly modified by the presence of spray at a mass
density less than 1% of air density.
Current research is focused on discovering whether
the relationships indicated above are likely to develop

further as wind speeds rise from the highest value analyzed by Fairall et al. (40 m s−1) toward those typical
31


FIG. 4. Prediction of individual tropical cyclone maximum intensity by applying the techniques of Holland (1997) and Emanuel (1991)
to Willis Island soundings applied to tropical cyclones in the eastern Australian region and within approximately 500 km of the island. The
data are plotted as a time series of maximum intensity predictions compared to the actual cyclone intensity change [adapted from Tonkin
(1996)].
32

Vol. 79, No. 1, January 1998


of intense tropical cyclones (50–60 m s−1). Initial efforts to achieve more comprehensive statistical modelling of droplet trajectories suggest that spray distribution may be critically dependent on a certain velocity ratio. It is when vertical gust components have a
standard deviation far greater than the terminal velocities of most droplets that it is anticipated that a substantial thickening of the spray zone will occur and
hence an enhancement of the effects of ocean spray on
tropical cyclone thermodynamics. This may be a significant restraining mechanism on any increase in the
maximum intensity of tropical cyclones in response
to increasing SST because the deep ocean spray zone
of intense tropical cyclones could constrain further
thermodynamic development.
Some recent studies by Gray (1996) show that
momentum considerations are a fundamental component that must also be taken into account in combination with lapse rates in upscaling models of MPI. As
a tropical cyclone’s maximum winds increase, its lowlevel frictional dissipation goes up with the square or
slightly higher power of the wind. Eyewall cloud
buoyancy, although always large in the weak stages
of the cyclone, rises at a much slower rate than does
friction. At high tropical cyclone intensities, eyewall
buoyancy is still only marginally (or not at all) enough
to support the larger increase in the need for eyewall

cloud vertical motion to balance friction. A point is
reached where the buoyancy-driven eyewall cloud
vertical motion is unable to increase sufficiently to
match the exponential rise in tropical cyclone momentum requirements. The frictional momentum dissipation was considered in the MPI approach of Emanuel
(1995) and his recent study (Emanuel 1997) further
shows that the MPI estimated by the thermodynamic
approach can be rederived from energetic considerations, and Holland (1997) showed the energetic consistency in his MPI estimation. Nevertheless, we still
lack a clear understanding of how the tropical cyclone
inner-core dynamics and thermodynamics limit its intensification.
b. Global climate model (GCM) studies of tropical
cyclone intensity
Haarsma et al. (1992) found an increase in the number of more intense tropical disturbances and in the
intensity of the most intense storms in a warmer GCM
climate. The increase of the maximum simulated wind
speed is about 20%. They also suggested that the GCM
severely restricts the maximum possible intensity of
the simulated tropical storms because of its coarse
Bulletin of the American Meteorological Society

spatial resolution. Bengtsson et al. (1995, 1996) found
that although the number of modeled storms is significantly reduced in their T106 GCM simulation, there
seems to be no reduction in their overall strength.
Li (1996) has applied the Emanuel and Holland
approaches to the climate models used in the MECCA
present-day and greenhouse intercomparison (Howe
and Henderson-Sellers 1997). He shows that the individual models produce widely varying values of MPI
for current climatic conditions, largely due to the poor
thermodynamic structure of the model atmosphere and
the poorly predicted SSTs. Li (1996) does find increased intensity of cyclones using both the Holland
and Emanuel thermodynamic models, although the

increases in MPI found in his analysis are inside the
uncertainty range derived from individual model predictions. This introduces considerable uncertainties for
direct application to climate change predictions and
calls into question the results of “downscaling” by
means of embedding mesoscale models into global
climate models (e.g., Walsh and Watterson 1997).
The coarse- and large-scale vortices generated by
GCMs do not capture the detailed core-region physical and dynamical processes that are known to be important to tropical cyclone intensification, including
oceanic coupling, and they do not have the capacity
to fully develop intense cyclones by the thermodynamic processes that feed back between the ocean and
the cyclone. Their usefulness as prediction tools depends upon the degree to which the cyclone intensity
is governed by external environmental factors, which
is not well known.
As pointed out by Gray, tropical cyclone potential
intensity (MPI) appears to depend on the existence of
background conditionally unstable lapse rates from the
surface to 300 hPa, or, equivalently, θe decreases from
the surface to 300 hPa, which is the usual level of
strongest tropical cyclone eyewall cloud updrafts.
Buoyancy decreases above this level. For example, the
northwest Pacific has the highest background value of
conditional instability from the surface to 300 hPa, and
the most intense tropical cyclones occur in this region.
Other regions with lower values of this quantity have
weaker or no cyclones. The question to be addressed
is how this surface to 300 hPa lapse rate will change
as global warming occurs. In recent years, a number
of studies has been done to investigate the moist stability and CAPE in the tropical atmosphere (e.g.,
Rennó and Ingersoll 1996; Robe and Emanuel 1996).
Rennó and Ingersoll (1996) argue that a necessary consequence of CO2-induced warming is larger CAPE,

33


because the mean temperatures at which radiation is
absorbed and emitted become more different. However, cloud feedbacks may reduce this effect.
Most global models indicate large mid- to uppertropospheric warming (3°–6°C) in their doubled CO2
simulations over the tropical oceans. They also show
only small 1°–2°C surface warming in the tropical cyclone ocean basins. Since surface moisture increases
occur with the surface warming, little or no background environmental surface to 300 hPa θe gradient
change is expected with global warming. Indeed, results from Bengtsson et al. (1996) did not show any
large change in the atmospheric moist stability in their
doubled CO2 simulation. From this point of view,
more intense tropical cyclones should not be expected
in a greenhouse-warmed climate.
By analyzing six MECCA model results, Li (1996)
shows large increases of atmospheric dry static stability (∂θ/∂p) in all MECCA models with larger warming in the high troposphere than at the surface (Fig. 5).
As noted by Holland (1997), this is quite different from
the destabilization that occurs when SST increases
each year in current climate (Fig. 5a). Li also finds that
the atmospheric moist static instability (∂θe/∂p) (calculated as the difference of θe between models’ surface and 200 hPa) exhibits only slight changes following greenhouse warming in the MECCA models. Further analysis from five of the MECCA models shows
that the instability of the low-level atmosphere is increased in the greenhouse-warmed climate but the stability is increased in the middle and upper troposphere.
Calculations of CAPE from these models show the
major increases to be limited to the low levels. This is
quite different from seasonal changes in current climate (Fig. 5b) where the instability increases through
a deep layer. How the atmospheric thermodynamic structure will change in the future climate and any implication this may have for changes in tropical cyclone
intensities needs to be addressed in future evaluations.
c. Distribution of tropical cyclone intensities
No information is available from current research
on changes in the distribution of cyclone intensities.
A net skewing of the intensity distribution up or down

could have a greater effect than changes in the worst
possible case. Landsea and Gray (1992) have found
climatic indicators for the gross distribution of hurricane intensities for the North Atlantic. Recent work
by Holland and collaborators has also found that there
may be environmental signatures in the Australian–
southwest Pacific region. However, such techniques
34

have not yet been applied successfully to climate simulations. DeMaria and Kaplan (1993, 1994) found that
the difference between current intensity and an empirically defined MPI provided a good predictor of
whether hurricanes in the North Atlantic would continue intensifying. This implies that a higher MPI will
lead to a greater frequency of intense cyclones in general, but this is not supported by the results in Fig. 4,
which indicate that there are substantial local and temporal variations of MPI that affect individual tropical
cyclones. It is concluded that there is insufficient evidence with which to predict changes in tropical cyclone intensity distribution.
5. Tropical cyclones in a greenhousewarmed world
The impacts on society by tropical cyclones have
been marked by a substantial decrease in deaths in the

FIG. 5. Simulated changes (enhanced greenhouse minus control
simulation) in potential temperature (a) and equivalent potential
temperature (b) over the GCM grid point near Willis Island (16.3°S,
149.97°E) in August for five MECCA models with observed
seasonal changes (February minus August) at the same location.
Vol. 79, No. 1, January 1998


developed nations, but a rapid increase of economic
damage and disruption of burgeoning coastal communities over the past few decades. The insurance industry in particular has experienced a rapid increase in
losses from tropical cyclone disasters during the last
decade. This has been caused, to a large extent, by increasing coastal populations, by increasing insured

values in coastal areas, and, perhaps, by a rising sensitivity of modern societies to disruptions of infrastructure. However, the insurance industry is worried about
the possibility of increasing frequencies and/or intensities of tropical cyclones in addition to the higher
exposures in coastal areas. Until scientific predictions
provide conclusive proof that these fears are unwarranted, the industry has to prepare itself for extreme
catastrophic losses by means of appropriate reserves
and restrictive underwriting.
Some progress has been made toward understanding the possible impacts on tropical cyclones of greenhouse warming. Detailed empirical and theoretical
studies have greatly improved our understanding of
what is not known and, therefore, which topics offer
the greatest likelihood of improved prediction skill.
These include
• increased realism in coupled ocean–atmosphere
global climate models;
• improved observations of air–sea interactions and
other aspects of tropical cyclone genesis and evolution; and
• further and more complete paleoclimatological
analyses relating past climate changes to changes
in tropical cyclone activity.

quency, area of occurrence, time of occurrence, mean
intensity or maximum intensity of tropical cyclones
will change” (Houghton et al. 1996, p. 334). We believe that it is now possible to improve on this statement. In particular:
• there is no evidence to suggest any major changes
in the area or global location of tropical cyclone
genesis in greenhouse conditions;
• thermodynamic “upscaling” models seem to have
some skill in predicting maximum potential intensity (MPI); and
• these thermodynamic schemes predict an increase
in MPI of 10%–20% for a doubled CO2 climate but
the known omissions (ocean spray, momentum restriction, and possibly also surface to 300 hPa lapse

rate changes) all act to reduce these increases.
Acknowledgments. The process used to generate this state-ofthe-art review extended from June 1996 to March 1997. The 10
members of the WMO/CAS/TMRP Committee (A. HendersonSellers, G. Berz, R. Elsberry, K. Emanuel, W. Gray, C. Landsea,
G. Holland, J. Lighthill, S.-L. Shieh, P. Webster) submitted upto-date assessments. These were synthesized into a single paper
by the rapporteur (Dr. H. Zhang) and the chairman (Professor A.
Henderson-Sellers). This draft was circulated to all committee
members and also reviewed by attendees at the ONR Symposium
on Tropical Cyclones in December 1996. Eleven scientists (K.
McGuffie, W. Gray, R. Elsberry, M. Lander, F. Wells, G. Holland,
J. Evans, L. Avila, I. Ginis, C. Landsea, and R. Abbey) reviewed
the document during a working session of the ONR symposium.
The resulting final version was circulated to all the committee
members for agreement. We are very grateful to all those who
participated in this process.

Appendix: List of Acronyms
Since the production of the 1996 IPCC reports, our
knowledge has advanced to permit the following summary.
• There are no discernible global trends in tropical cyclone number, intensity, or location from historical
data analyses.
• Regional variability, which is very large, is being
quantified slowly by a variety of methods.
• Empirical methods do not have skill when applied
to tropical cyclones in greenhouse conditions.
• Global and mesoscale model-based predictions for
tropical cyclones in greenhouse conditions have not
yet demonstrated prediction skill.
The IPCC “Science of Climate Change” report
stated that “it is not possible to say whether the freBulletin of the American Meteorological Society

AGCM
CAPE
CAS
CLIVAR

Atmospheric General Circulation Model
Convective available potential energy
Commission for Atmospheric Sciences
Climate Variability and Predictability
Programme, WCRP
ECMWF European Centre for Medium-Range
Weather Forecasts
ENSO
El Niño–Southern Oscillation
GCM
General circulation model or global
climate model
GFDL
Geophysical Fluid Dynamics Laboratory
ICSU
International Council of Scientific
Unions
IPCC
Intergovernmental Panel on Climate
Change
ITCZ
Intertropical Convergence Zone
35

MECCA Model Evaluation Consortium for
Climate Assessment
MLO
Mixed-Layer Ocean (model)
MPI
Maximum potential intensity
NCDC
National Climate Data Center
OAGCM (coupled) Ocean–Atmosphere General
Circulation Model
QBO
Quasi-biennial oscillation
SGP
Seasonal genesis parameters
SST
Sea surface temperature
TC
Tropical cyclone
TMRP
Tropical Meteorology Research
Programme
WCRP
World Climate Research Programme
WMO
World Meteorological Organization
References
Anthes, R. A., 1982: Tropical Cyclones: Their Evolution, Structure
and Effects. Meteor. Monogr., No. 41, Amer. Meteor. Soc., 208 pp.
Baik, J.-J., M. DeMaria, and S. Raman, 1990: Tropical cyclone
simulations with the Betts convective adjustment scheme. Part

II: Sensitivity experiments. Mon. Wea. Rev., 118, 529–541.
Bengtsson, L., M. Botzet, and M. Esch, 1994: Hurricane-type
vortices in a general circulation model, Part I. Tech. Rep. No.
123, Max-Planck-Institut für Meteorologie, Hamburg, 42 pp.
[Available from Max-Planck-Institut für Meteorologie,
Bundestr. 55, 2000 Hamburg 13, Germany.]
——, ——, and ——, 1995: Hurricane type vortices in a general
circulation model. Tellus, 47A, 175–196.
——, ——, and ——, 1996: Will greenhouse gas-induced warming over the next 50 years lead to higher frequency and greater
intensity of hurricanes? Tellus, 48A, 57–73.
Black, P. G., 1993: Evolution of maximum wind estimates in typhoons. Tropical Cyclone Disasters, J. Lighthill, Z. Zhemin,
G. Holland, and K. Emanuel, Eds., Peking University Press,
104–115.
Broccoli, A. J., and S. Manabe, 1990: Can existing climate models be used to study anthropogenic changes in tropical cyclone
climate? Geophys. Res. Lett., 17, 1917–1920.
Byers, H. R., 1944: General Meteorology. 2d ed. McGraw-Hill,
645 pp.
Chan, J. C. L., 1985: Tropical cyclone activity in the northwest
Pacific in relation to the El Niño/Southern Oscillation phenomenon. Mon. Wea. Rev., 113, 599–606.
——, and J. Shi, 1996: Long-term trends and interannual variability in tropical cyclone activity over the western North Pacific.
Geophys. Res. Lett., 23, 2765–2767.
Chang, J.-C. J., 1996: An exploratory study of the relationship between annual frequency of invaded typhoons in Taiwan and
El Niño/Southern Oscillation. Terr., Atmos. Oceanic Sci., 7,
83–105.
Cubasch, U., K. Hasselmann, H. Höck, E. Maier-Reimer, U.
Mikolajewicz, B. D. Santer, and R. Sausen, 1992: Time-dependent greenhouse warming computations with a coupled
ocean–atmosphere model. Climate Dyn., 8, 55–69.

36

DeMaria, M., and J. Kaplan, 1993: Verification of a statistical
hurricane intensity prediction model. Preprints, 20th Conf. on
Hurricanes and Tropical Meteorology, San Antonio, TX,
Amer. Meteor. Soc., 303–304.
——, and ——, 1994: Sea surface temperature and the maximum
intensity of Atlantic tropical cyclones. J. Climate, 7, 1324–
1334.
Diaz, H. F., and R. S. Pulwarty, Eds., 1997: Hurricanes: Climate
and Socioeconomic Impacts. Springer-Verlag, 292 pp.
Dong, K., 1988: El Niño and tropical cyclone frequency in the
Australian region and the Northwest Pacific. Aust. Meteor.
Mag., 36, 219–225.
——, and G. J. Holland, 1994: A global view of the relationship
between ENSO and tropical cyclone frequencies. Acta Meteor.
Sinica, 8, 19–29.
Drury, S., and J. L. Evans, 1993: Sea surface temperature and
CAPE: Importance for tropical cyclone intensity. Preprints,
20th Conf. on Hurricanes and Tropical Meteorology, San
Antonio, TX, Amer. Meteor. Soc., 89–92.
Dvorak, V. F., 1984: Tropical cyclone intensity analysis using
satellite data. NOAA Tech. Rep. NESDIS 11, 47 pp.
Emanuel, K. A., 1986: An air–sea interaction theory for tropical
cyclones. Part 1: Steady-state maintenance. J. Atmos. Sci., 43,
585–604.
——, 1987: The dependence of hurricane intensity on climate.
Nature, 326, 483–485.
——, 1991: The theory of hurricanes. Annu. Rev. Fluid. Mech.,
23, 179–196.
——, 1995: Sensitivity of tropical cyclones to surface exchange
coefficients and a revised steady-state model incorporating eye

dynamics. J. Atmos. Sci., 52, 3969–3976.
——, 1997: Some aspects of hurricane inner-core dynamics and
energetics. J. Atmos. Sci., 54, 1014–1026.
Evans, J. L., 1990: Envisaged impacts of enhanced greenhouse
warming on tropical cyclones in the Australian region. CSIRO
Division of Atmospheric Research Tech. Paper, 20, 31 pp.
[Available from Division of Atmospheric Research, Private
Bag 1, Mordialloc, VIC 3195, Australia.]
——, 1992: Comment on “Can existing climate models be used
to study anthropogenic changes in tropical cyclone climate?”
Geophys. Res. Lett., 19, 1523–1524.
——, 1993: Sensitivity of tropical cyclone intensity to sea surface temperature. J. Climate., 6, 1133–1140.
——, and R. J. Allan, 1992: El Niño/Southern Oscillation modification to the structure of the monsoon and tropical cyclone
activity in the Australian region. Int. J. Climatol., 12, 611–623.
——, B. F. Ryan, and J. L. McGregor, 1994: A numerical exploration of the sensitivity of tropical cyclone rainfall intensity to
sea surface temperature. J. Climate., 7, 616–623.
Fairall, C. W., J. D. Kepert, and G. J. Holland, 1994: The effect
of sea spray on surface energy transports over the ocean. Global Atmos. Ocean Syst., 2, 121–142.
Frank, W. M., 1987: Tropical cyclone formation. A Global View
of Tropical Cyclones, R. L. Elsberry, Ed., U.S. Office of Naval Research, 53–90.
Gray, S. L., 1994: Theory of mature tropical cyclones: A comparison between Kleinschmidt (1951) and Emanuel (1986).
JCMM Tech. Rep. 40, 50 pp. [Available from Joint Centre for
Mesoscale Meteorology, University of Reading, P.O. Box 240,
Reading, Berkshire RG6 2FN, United Kingdom.]

Vol. 79, No. 1, January 1998


Gray, W. M., 1968: Global view of the origin of tropical disturbances and storms. Mon. Wea. Rev., 96, 669–700.
——, 1975: Tropical cyclone genesis. Dept. of Atmos. Sci. Paper No. 234, Colorado State University, Fort Collins, CO, 121

pp. [Available from Dept. of Atmospheric Science, Colorado
State University, Ft. Collins, CO 80523.]
——, 1979: Hurricanes: Their formation, structure and likely role
in the general circulation. Meteorology over the Tropical Oceans,
D. B. Shaw, Ed., Royal Meteorological Society, 155–218.
——, 1981: Recent Advances in Tropical Cyclone Research from
Rawinsonde Composite Analysis. World Meteorological Organization, 407 pp.
——, 1984a: Atlantic seasonal hurricane frequency. Part I: El
Niño and 30 mb quasi-biennial oscillation influences. Mon.
Wea. Rev., 112, 1649–1668.
——, 1984b: Atlantic seasonal hurricane frequency. Part II: Forecasting its variability. Mon. Wea. Rev., 112, 1669–1683.
——, 1996: Limiting influences on the typhoon’s maximum potential intensity. ONR Tropical Cyclone Symp., Melbourne,
Australia, BMRC, 82–97.
——, and C. W. Landsea, 1992: African rainfall as a precursor of
hurricane-related destruction on the U.S. east coast. Bull. Amer.
Meteor. Soc., 73, 1352–1364.
——, ——, P. W. Mielke Jr., and K. J. Berry, 1993: Predicting
Atlantic seasonal tropical cyclone activity by 1 August. Wea.
Forecasting, 8, 73–86.
——, ——, ——, and ——, 1994: Predicting Atlantic seasonal
tropical cyclone activity by 1 June. Wea. Forecasting, 9, 103–
115.
Haarsma, R. J., J. F. B. Mitchell, and C. A. Senior, 1992: Tropical disturbances in a GCM. Climate Dyn., 8, 247–257.
Hayne, M., 1996: Northeast Australian cyclonic variability during
the late holocene. 3d Workshop, Tropical Cyclone Coastal Impacts Program, Brisbane, Australia, Bureau of Meteorology, 13.
Hebert, P. J., J. D. Jarrell, and M. Mayfield, 1996: The deadliest,
costliest, and most intense United States hurricanes of this
century (and other frequently requested hurricane facts).
NOAA Tech. Memo, NWS TPC-1, Miami, FL, 75–104.
Holland, G. J., 1981: On quality of the Australian tropical cyclone

data base. Aust. Meteor. Mag., 29, 169–181.
——, 1984a: On the climatology and structure of tropical cyclones
in the Australian/Southwest Pacific region, I: Data and tropical storms. Aust. Meteor. Mag., 32, 1–16.
——, 1984b: On the climatology and structure of tropical cyclones
in the Australian/Southwest Pacific region, II: Hurricanes.
Aust. Meteor. Mag., 32, 17–32.
——, Ed., 1993: The Global Guide to Tropical Cyclone Forecasting. World Meteorological Organization, 337 pp.
——, 1995: Scale interaction in the western Pacific monsoon.
Meteor. Atmos. Phys., 56, 57–79.
——, 1997: The maximum potential intensity of tropical cyclones.
J. Atmos. Sci., 54, 2519–2541.
——, J. L. McBride, and N. N. Nicholls, 1988: Australian tropical cyclones and the Greenhouse effect. Greenhouse: Planning for Climate Change, G. I. Pearman, Ed., CSIRO, 438–455.
Houghton, J. T., L. G. Meira Filho, B. A. Callander, N. Harris, A.
Kattenberg, and K. Maskell, 1996: Climate Change 1995: The
Science of Climate Change. Contribution of Working Group I
to the Second Assessment of the Intergovernmental Panel on
Climate Change. Cambridge University Press, 572 pp.

Bulletin of the American Meteorological Society

Howe, W., and A. Henderson-Sellers, Eds., 1997: Assessing Climate Change: Results from the Model Evaluation Consortium
for Climate Assessment. Gordon and Breach, 422 pp.
Keen, T. R., and R. L. Slingerland, 1993: Four storm-event beds
and the tropical cyclones that produced them: A numerical
hindcast. J. Sed. Petroe., 63, 218.
Kleinschmidt, E., Jr., 1951: Grundlagen einer Theorie des
tropischen Zyklonen. Arch. Meteor. Geophys. Bioklimatol.,
A4, 53–72.
Lander, M. A., 1994: An exploratory analysis of the relationship
between tropical storm formation in the western North Pacific

and ENSO. Mon. Wea. Rev., 122, 636–651.
Landsea, C. W., 1993: A climatology of intense (or major) Atlantic hurricanes. Mon. Wea. Rev., 121, 1703–1713.
——, and W. M. Gray, 1992: The strong association between
western Sahel monsoon rainfall and intense Atlantic hurricanes. J. Climate, 5, 435–453.
——, N. Nicholls, W. M. Gray, and L. A. Avila, 1996: Downward trends in the frequency of intense Atlantic hurricanes
during the past five decades. Geophys. Res. Lett., 23, 1697–
1700.
Lawrence, J. R., and S. D. Gedzelman, 1996: Low stable isotope
ratios of tropical cyclone rains. Geophys. Res. Lett., 23, 527–
530.
Li, S., 1996: Estimating tropical cyclone intensity from climate
models. M.S. thesis, Climatic Impacts Centre, Macquarie University, Sydney, Australia, 136 pp.
Lighthill, J., G. J. Holland, W. M. Gray, C. Landsea, K. Emanuel,
G. Craig, J. Evans, Y. Kurihara, and C. P. Guard, 1994: Global climate change and tropical cyclones. Bull. Amer. Meteor.
Soc., 75, 2147–2157.
Liu, K.-B., and M. L. Fearn, 1993: Lake-sediment record of late
Holocene hurricane activities from coastal Alabama. Geology,
21, 793–796.
Malkus, J. S., and H. Riehl, 1960: On the dynamics and energy
transformations in steady-state hurricanes. Tellus, 12, 1–20.
McBride, J. L., 1981: Observational analyses of tropical cyclone
formation. Part I: Basic description of data sets. J. Atmos. Sci.,
38, 1117–1131.
——, 1995: Tropical cyclone formation. Global Perspectives on
Tropical Cyclones, R. L. Elsberry, Ed., World Meteorological Organization, 63–105.
Neumann, C. J., 1993: Global overview. Global Guide to Tropical Cyclone Forecasting, WMO/TC-No. 560, Rep. TCP-31,
World Meteorological Organization, Geneva, Switzerland,
220 pp. [Available from World Meteorological Organization,
P.O. Box 2300, 41, Avenue Giuseppe Motta, CH-1211,
Geneva 2, Switzerland.]

——, G. W. Cry, F. L. Caso, and B. R. Jarvinen, 1985: Tropical
Cyclones of the North Atlantic Ocean, 1871–1980. NOAA
Special Publication, 174 pp.
Nicholls, N., 1984: The Southern Oscillation, sea surface temperature, and interannual fluctuations in Australian tropical cyclone
activity. J. Climatol, 4, 661–670.
——, 1992: Recent performance of a method for forecasting Australian seasonal tropical cyclone activity. Aust. Meteor. Mag.,
40, 105–110.
Pudov, V. D., and S. A. Petrichenko, 1988: Temperature of the
South China Sea surface and tropical cyclones. Meteorologiya
i Gidrologiya, 12, 327–347.

37


Raper, S. C. B., 1993: Observational data on the relationships between climatic change and the frequency and magnitude of severe tropical storms. Climate and Sea Level Change: Observations, Projections and Implications, R. A. Warrick, E. M.
Barrow, and T. M. L. Wigley, Eds., Cambridge University
Press, 192–212.
Rennó, N. O., and A. P. Ingersoll, 1996: Natural convections as
a heat engine: A theory for CAPE. J. Atmos. Sci., 53, 572–
585.
Revell, C. G., and S. W. Goulter, 1986: South Pacific tropical cyclones and the Southern Oscillation. Mon. Wea. Rev., 114,
1138–1145.
Riehl, H., 1954: Tropical Meteorology. McGraw-Hill, 392 pp.
Robe, F. R., and K. A. Emanuel, 1996: Moist convective scaling:
Some inferences from three-dimensional cloud ensemble simulations. J. Atmos. Sci., 53, 3265–3275.
Ryan, B. F., I. G. Watterson, and J. L. Evans, 1992: Tropical cyclone frequencies inferred from Gray’s Yearly Genesis Parameter: Validation of GCM tropical climates. Geophys. Res. Lett.,
19, 1831–1834.
Shapiro, L. J., 1987: Month-to-month variability of the Atlantic
tropical circulation and its relationship to tropical cyclone formation. Mon. Wea. Rev., 115, 2598–2614.
Simpson, R. H., 1974: The hurricane disaster potential scale.

Weatherwise, 27, 169–186.
Tonkin, H., 1996: Investigating thermodynamic models of tropical cyclone intensity. M.S. thesis, Climatic Impacts Centre,
Macquarie University, Sydney, Australia, 127 pp.

38

——, C. Landsea, G. J. Holland, and S. Li, 1997: Tropical cyclones and climate change: A preliminary assessment. Assessing Climate Change: Results from the Model Evaluation Consortium for Climate Assessment, W. Howe and A. HendersonSellers, Eds., Gordon and Breach, 327–360.
Walsh, K., and I. G. Watterson, 1997: Tropical cyclone-like vortices in a limited area model: Comparison with observed climatology. J. Climate, 10, 2240–2259.
Watson, R. T., M. C. Zinyoweya, and R. H. Moss, Eds., 1996:
Climate Change 1995: Impacts, Adaptations and Mitigation
of Climate Change: Scientific-Technical Analyses. Contribution of Working Group II to the Second Assessment Report of
the Intergovernmental Panel on Climate Change. Cambridge
University Press, 879 pp.
Watterson, I. G., J. L. Evans, and B. F. Ryan, 1995: Seasonal and
interannual variability of tropical cyclogenesis: Diagnostics
from large-scale fields. J. Climate, 8, 3052–3066.
WMO, 1994: Proceedings of the Third WMO/ICSU International
Workshop on Tropical Cyclones (IWTC-III). World Meteorological Organization, 41 pp.
Xue, Z., and C. J. Neumann, 1984: Frequency and motion of western North Pacific tropical cyclones. Tech. Memo. NWS NHC
23, NOAA, 80 pp. [Available from NOAA/AOML Hurricane
Research Division, Rickenbacker Causeway, Miami, FL
33149.]

Vol. 79, No. 1, January 1998